基于BP神经网络的Falcon分选电子废弃物预测模型  

BP Neural Network Prediction Model for Sorting Electric and Electronic Waste by Falcon

作　　者：张瑾[1] 赵跃民[2] 

机构地区：[1]中国矿业大学环境与测绘学院 [2]中国矿业大学化工学院,江苏徐州221008

出　　处：《环境科学与技术》2013年第3期74-77,共4页Environmental Science & Technology

基　　金：国家自然科学基金创新研究群体项目(50921002);江苏省基础研究计划-创新学者攀登项目(BK2010002);中国博士后科学基金特别资助项目(201003607);高等学校博士学科点新教师专项科研基金(20100095120003);液固流态化分选细粒物料的非线性动力学研究;江苏省自然科学基金(BK2012136)

摘　　要：Falcon分选电子废弃物的影响因素主要有水压、转动频率、入料浓度,为了进一步研究影响因素与分选效果的关系,文章利用Design-Expert 7.1软件设计出三因素三水平的实验条件,利用Falcon分选得出数据。将实验数据和BP神经网络相结合,将影响因素作为神经网络的输入,品位和产率作为输出,经过BP训练后得到输入与输出的关系。对实验影响因素与分选效果的关系进行分析,结果与实际情况比较吻合。运用MATLAB实现BP神经网络仿真,仿真结果与最小二乘法下的结果相比较误差较小,输出向量与实际实验结果接近。Water pressure,rotation frequency and feed concentration are major influencing factors of Falcon sorting electronic waste.In order to study the further relationship between factors and separation effect,Design-Expert 7.1 software was used to design 3 factors and 3 levels experimental conditions,and data were derived by using Falcon sorting.Experimental data and BP neural network were combined.With the influencing factors as the input of neural network,quality and productivity as the output,the data were trained to get the relationship between input and output by BP neural network.The relationship between influences of experimental factors and separation efficiency was analyzed,and the results were consistent with the reality.Use of MATLAB to achieve BP neural network simulation,the simulation results and the least squares method were compared,with small error and close of the output vector and actual experimental results.

关 键 词：BP神经网络 FalconSB40 电子废弃物 最小二乘法 MATLAB 

分 类 号：X76[环境科学与工程—环境工程]